Multilevel least-change Newton-like methods for equality constrained optimization problems |
| |
Authors: | S M Grzegórski |
| |
Institution: | (1) Numerical Analysis Department, Marie Curie-Sklodowska University, 20-031 Lublin, Poland |
| |
Abstract: | This paper deals with the application of multilevel least-change Newton-like methods for solving twice continuously differentiable
equality constrained optimization problems. We define multilevel partial-inverse least-change updates, multilevel least-change
Newton-like methods without derivatives and multilevel projections of fragments of the matrix for Newton-like methods without
derivatives. Local andq-superlinear convergence of these methods is proved. The theorems here also imply local andq-superlinear convergence of many standard Newton-like methods for nonconstrained and equality constraine optimization problems. |
| |
Keywords: | Nonlinear programming augmented Lagrangian method Newton-like methods least-change updates multilevel orthogonal projections q-superlinear convergence |
本文献已被 SpringerLink 等数据库收录! |